What To Do About Crime's Persistent Perception Problem?

A new-ish poll and some ideas for the future.

YouGov released a poll recently asking two questions on crime:

1. Since 1990, would you say murder rates in U.S. cities have...?

And

2. Since 2020, would you say murder rates in U.S. cities have...?

This is a variation on a question that Gallup has asked for decades, but I like it a lot more because it’s very specific to murder over set timeframes while Gallup’s question asks about a more vaguely defined “crime”.

It won’t shock you to learn that a majority of respondents answered both questions with “increased” by fairly wide margins.

Adding things up, that’s 49 percent saying murder increased since 2020 compared to 17 percent saying it decreased (35 percent said it either stayed about the same or they weren’t sure). Turning to the question about 1990 we get a similarly sized gap:

Murder has quite obviously fallen a lot since 2020 and it’s fallen a miraculous amount since 1990. If you want to take the question literally and only look at data from cities then you get a murder rate of 11.5 per 100k in US cities in 1990 and a rate of 6.8 per 100k in 2023 (the 2024 data will almost certainly show a sizable decline when it’s released in a few weeks).

Philip Bump wrote up a good summary of the survey and I don’t need to rehash his discussion of why so many respondents are so wrong. John Roman also had a good write up of the issue that I’d recommend reading.

I don’t want to rehash the reasons that I think most prominently explain why peoples’ perceptions of crime differ so strongly from reality, I wrote about that a year and a half ago. Hell, we could go back a century and find police chiefs using public perceptions of crime waves as their impetus to start collecting standardized crime data in the first place.

Instead, I want to discuss why the YouGov poll provides some optimism that the perception gap can be closed and the steps my team are taking and plan to take to do so.

Let’s start with the optimism which, in my opinion, comes in the form of the poll’s breakdown by age.

An insane 63 percent of people 65+ in this poll said that murder had increased since 1990 . Someone who is 65 now was 30 in 1990. They have lived through crime at its very worst in the early 1990s and then twice seen remarkable drops, yet nearly two-thirds of respondents at that age said murder was higher now than back then. They may be wrong but at least they’re confidently wrong with only 8 percent saying they aren’t sure whether it increased.

On the other end of the spectrum (though not far enough over to get the question right) were people 18 to 29 and people of the “good” generation (Millennials aged 30 to 44). Less than half of those groups said murder had increased since 1990 and 2020 and a decent sized chunk of each group said they weren’t sure.

Younger groups do not believe murder is higher than it was in 1990 as much as older groups. They also are more willing to say they aren’t sure than older groups. I obviously could be reading too much into a single poll’s results, but these points imply to me that younger generations are at least open to the idea of closing the perception-to-reality gap.

The real question, therefore, is how to start to close the gap. I posted my initial reaction to seeing the YouGov poll the other day and many of the replies focused on the role of the media in leading people to think crime is rising.

But I don’t think it’s that simple. My parents are in their 70s and they watch the local news, but it’s not like younger generations are watching the 6 PM news each night to determine whether crime is rising or falling where they are.

There has also been a decent level of media coverage of murder’s decline in 2025. Within the span of a week in late July — around when th3 survey was running — we got articles on it from the Washington Post, New York Times, Wall Street Journal, Financial Times, and Bloomberg. FBI Director Kash Patel went on the Joe Rogan podcast in June and discussed a 20 percent drop in murder to potentially the lowest level ever recorded, Deputy Director Dan Bongino has been tweeting about it, even John Lott is discussing it at The Federalist. Even President Trump has talked about enormous drops in murder nationally when not decrying surging crime in DC and elsewhere.

So “the media” is definitely talking about the murder decline, yet these pesky perceptions persist…problematically. The extreme politicization of the issue may be new, but the basic problem is isn’t new. People have long told Gallup that crime has risen in the last year nationally regardless of what the data tells us.

I know that this exact political moment is somewhat unique in terms of perceptions of crime trends, but these problems extend far beyond this year, or even this century. Standardized crime statistics began being collected — at least in part — because police chiefs wanted to be able to respond to people assuming that we were in the midst ofa crime wave.

I’m certainly no communications expert, but it’s my hope that at least part of the problem is that the platforms being used to spread information on crime trends are not the platforms younger people are using to receive news.

Most people get news from their “digital devices” and people are far more likely to get news from YouTube, Instagram, or TikTok than they are to get it from Twitter. All of which is why we are launching efforts on those platforms to provide content on crime data and trends and hopefully reach a wider audience that is not reading this Substack.

To that end, we created the Aggregated YouTube series. The first video is below for you to enjoy and by all means please head over there to subscribe. The goal is to produce new videos every other week covering current crime trends, looks at individual cities, and other fun quirks that pop up. Here’s the first video covering a brief history of crime data:

And the second which is all about the Real-Time Crime Index:

The third video looked at crime trends at midyear:

The fourth video that came out last week talked about how posting on TikTok led to a surge in auto thefts:

And, very excitedly, I’m creating a new podcast called (drumroll) the Jeff-alytics Podcast! I’m pumped about this and will have more information in a few weeks, but the goal is to talk to journalists, academics, policymakers, philanthropists, and more with the specific goals of 1) better understanding the gap between crime perceptions and the actual data and 2) exploring what is being done to bridge that gap. It’s my hope that it’s an expansive conversation about crime and criminal justice and how we can better communicate what data and research finds to a public that typically only thinks things are getting worse (sometimes they’re right!).

Like I said, I’ll have more about the podcast in a few weeks, but the goal is to launch sometime in late-September and do for crime data and trends what Mike Duncan did for teaching people to overthrow regimes in protest of unfair taxes. People seem to like podcasts, hopefully they’ll like this one too.

So that’s the plan. It’s obviously quite a steep hill to climb, but providing content to people on the platforms they use is hopefully one way to close the gap between perceptions and reality.